Create powerful dataviz with R

Ilya Kashnitsky

14 June 2021

before we start

https://twitter.com/RuJEconomics/status/1191697644499984384

https://www.visualcapitalist.com/wp-content/uploads/2018/03/money-happiness-large.html

materials for the course

https://github.com/ikashnitsky/dataviz-mpidr

rule 0 – DO VISUALIZE YOUR DATA

Anscombe’s Quartet

http://i.imgur.com/QA3Ss8D.png

New #chi2017 paper is up. Don't trust statistics alone, visualize your data!https://t.co/amnbAYvsq1 pic.twitter.com/1s6vkge6dl
— Justin Matejka (@JustinMatejka) May 1, 2017

Visualizing summary statistics in a #boxplot is great. But don't forget to explore the underlying data distribution.

Thanks to @JustinMatejka for this great educational dataset.#30DayChartChallenge | #Day27 | #educational #dataviz #Rstats #ggplot2 #statistics #violinplot pic.twitter.com/JfNwMzuash
— Mihaela Bozukova (@MBozukova) April 27, 2021

https://i.imgur.com/Ym0oKHj.png

https://i.imgur.com/T7xz3UN.png

https://i.imgur.com/MxUNZxV.png

Remade the @WSJ measles vaccination visualization using ggplot2 and the animation package #rstats pic.twitter.com/cnIScD9ymI
— Michael Lee 🚴📊🇷 (@mikeleeco) June 19, 2017

rule 1 – text should be horizontal

Day 4 #30DayChartChallenge Magical

My choice is to show an easy 🪄magic🪄 trick to routinely improve the arguably most popular plot type

The #dataviz rule is simple:
👉text in plots should be horizontal👈

A related (not perfectly articulated) blog post: https://t.co/M2yXspYipc pic.twitter.com/nhQAVIMTfc
— Ilya Kashnitsky (@ikashnitsky) April 4, 2021

https://ikashnitsky.github.io/2019/dotplot/

example of a figure improvement

https://doi.org/10/ggbtpx

https://gist.github.com/ikashnitsky/2800295e304b4858be553432de4a0d11

https://twitter.com/ikashnitsky/status/1192901559568523269

rule 2 – on slides, text should be as large as possible

[side note]
It's even more important than the ones coming right after:
– #RotateTheDamnPlot
– #IncreaseTheDamnFont pic.twitter.com/LA2kMrPbLh
— Ilya Kashnitsky (@ikashnitsky) April 8, 2021

https://i.imgur.com/u1nCiJt.jpg

https://i.imgur.com/bYePASN.jpg

https://i.imgur.com/nvNEc2F.jpg

https://i.imgur.com/KpBVKP1.png

rule 3 – mind colors, especially regarding colorblind friendliness

https://t.co/KPqzPh06Iq pic.twitter.com/P1YWTIWooR
— Ilya Kashnitsky (@ikashnitsky) May 6, 2020

Great news #rstats #dataviz — the default go-to color selection just got richer 💥🎉

Knowing where it comes from, I know better than to question wherever “turbo” color scheme is really colorblind friendly, but it really blows my mind that something of these colors can be… https://t.co/AVo6cdMD32 pic.twitter.com/pIud4NRjhY
— Ilya Kashnitsky (@ikashnitsky) April 12, 2021

rule 4 – highlight what’s important for the story

Spent way too long trying to work out what was so special about her knee https://t.co/0epzaJeQKk
— Alex Selby-Boothroyd (@AlexSelbyB) June 13, 2021

https://barcanumbers.wordpress.com/2018/12/06

https://www.ft.com/content/a26fbf7e-48f8-11ea-aeb3-955839e06441

rule 5 – plots don’t have to be overly complicated to be powerful

Climate change is happening now & is happening everywhere. Our lives are already being affected by the warming of the planet.

Human activities are causing these fundamental changes to the climate. Burning fossil fuels & deforestation are the primary reasons.

Our choices matter. pic.twitter.com/SPnkVYq10u
— Ed Hawkins (@ed_hawkins) April 9, 2021

Austria vs Germany ⚽️ The First 111 Years #AUTGER international #football results since 1907:
🇦🇹 10 wins
🇩🇪 25 wins
✖️ 6 draws#dataviz by @tinafrank & @metropop_eu pic.twitter.com/iRjXmpNsOe
— ramon bauer (@metropop_eu) June 3, 2018

https://fivethirtyeight.com/features/why-the-oldest-person-in-the-world-keeps-dying

a great example

Friday and don't know what to do? Train your #ggplot2 skills! 📊
My first blog post “The Evolution of a ggplot (Ep. 1)” is online 🎉https://t.co/fWG4PmyyhU

Thanks to @R4DScommunity @rstudio @dataandme for all the neat #rstats tricks! #tidyverse #DataViz #TidyTuesday #R4DS pic.twitter.com/RRMTSjXFgj
— Cédric Scherer (@CedScherer) May 17, 2019

(not a rule) suggestion

When possible and meaningful for you story – animate

⚽️ With the #EURO2020 approaching, you've probably all been wondering what ~1 million of passes would looks like if all shot at the same time, right? Well wonder no more…

Thanks @StatsBomb for making the data openly available! @observablehq notebook https://t.co/dO4sfQpRi4 pic.twitter.com/fqbdK6PjRd
— Karim Douïeb (@karim_douieb) June 9, 2021

https://www.ft.com/video/83703ffe-cd5c-4591-9b4f-a3c087aa6d19

Or make it completely interactive

https://jschoeley.shinyapps.io/hmdexp

Dataviz principles

https://policyviz.com/2018/08/07/dataviz-cheatsheet/

https://ft-interactive.github.io/visual-vocabulary/

https://serialmentor.com/dataviz/

http://socviz.co/index.html

Jonas Shoeley’s slides

https://github.com/jschoeley/idem_viz

Tidyverse

pic.twitter.com/T4QBJaBD7s
— Trust me, I'm a “Biologist” (@TrustBiologist) June 13, 2021

The most influential R developer

Hadley Wickham

hadley

tidyverse

https://blog.rstudio.org/2016/09/15/tidyverse-1-0-0/

https://www.tidyverse.org/

.@rebeccmart and I are excited to welcome our new #rstats programmer to the family. Thanks to @create_self and @LanderAnalytics for the awesome baby themed hex shirts. #rstatsnyc #nycdatamafia pic.twitter.com/FqzUyDVEDo
— Jared Lander (@jaredlander) May 10, 2019

tidy data

Wickham, H. (2014). Tidy Data. Journal of Statistical Software, 59(10). Retrieved from http://www.jstatsoft.org/v59/i10

Tidy data is a standard way of mapping the meaning of a dataset to its structure.

A dataset is messy or tidy depending on how rows, columns and tables are matched up with observations, variables and types.

In tidy data:

Each variable forms a column.
Each observation forms a row.
Each type of observational unit forms a table.

On pipes

I used this idea today for teaching #rstats pipes and 🤞hopefully it helped? 🤞 https://t.co/XnpsIw96Hw pic.twitter.com/6sAmM4FDRy
— Andrew Heiss, PhD (@andrewheiss) September 16, 2019

Examples and exercises

Please follow me on the R script `"tidy.R"`

Visualizing data with `ggplot2`

A bit more motivation

https://graphics.reuters.com/SOCCER-EURO/yzdvxmjjnpx/

https://www.granvillematheson.com/post/self-portrait

I’d argue that by rotating the axis we change the story. If the axes were squared we’d just say Lewendowski is taking less shots but of a higher quality than Mbappé, but rotating may show Mbappé is less risk averse and it’s paying off https://t.co/FudNJC43P3
— Peter McKeever Ô_ō (@petermckeever) October 8, 2019

wowowowowowowow @OptaPro @Arsenal. In 18/19, only outshined by Messi in top 5 leagues for attacking contribution after progressive carries. Unbelievable signing. Look at how far he carried the ball before those shots. pic.twitter.com/HAehnA2Rlj
— Peter McKeever Ô_ō (@petermckeever) August 5, 2019

Plotting systems in R?

base
lattice
ggplot2

“The winner takes it all”

cat

Strength of `base` plotting system

Usually, base knows how to plot an object
Extremely easy to use if you are happy with the default settings
BUT
Now ggplot2 has the autoplot function

What makes `ggplot2` special?

“gg” means “Grammar of graphics”

http://www.springer.com/us/book/9780387245447

Extremely big and helpful community

Help
Examples
Rapid development
Extensions

ggplot2 resources

https://ggplot2.tidyverse.org/reference/

https://exts.ggplot2.tidyverse.org

https://pkg.garrickadenbuie.com/gentle-ggplot2/#1

https://emitanaka.org/workshopUTokyo2018/day1-session02-datavis.html#1

https://www.r-graph-gallery.com/cartogram.html

One last hint before we start coding

https://github.com/dreamRs/esquisse

Create powerful dataviz with R

Ilya Kashnitsky

14 June 2021

before we start

materials for the course

rule 0 – DO VISUALIZE YOUR DATA

Anscombe’s Quartet

rule 1 – text should be horizontal

example of a figure improvement

rule 2 – on slides, text should be as large as possible

rule 3 – mind colors, especially regarding colorblind friendliness

rule 4 – highlight what’s important for the story

rule 5 – plots don’t have to be overly complicated to be powerful

a great example

(not a rule) suggestion

When possible and meaningful for you story – animate

Or make it completely interactive

Dataviz principles

Jonas Shoeley’s slides

Tidyverse

The most influential R developer

Hadley Wickham

tidyverse

tidy data

On pipes

Examples and exercises

Please follow me on the R script `"tidy.R"`

Visualizing data with `ggplot2`

A bit more motivation

More

Plotting systems in R?

“The winner takes it all”

Strength of `base` plotting system

What makes `ggplot2` special?

Extremely big and helpful community

ggplot2 resources

One last hint before we start coding

`ggplot2` show

Please follow me on “`ggplot2.R`”

Create powerful dataviz with R

Ilya Kashnitsky

14 June 2021

before we start

materials for the course

rule 0 – DO VISUALIZE YOUR DATA

Anscombe’s Quartet

rule 1 – text should be horizontal

example of a figure improvement

rule 2 – on slides, text should be as large as possible

rule 3 – mind colors, especially regarding colorblind friendliness

rule 4 – highlight what’s important for the story

rule 5 – plots don’t have to be overly complicated to be powerful

a great example

(not a rule) suggestion

When possible and meaningful for you story – animate

Or make it completely interactive

Dataviz principles

Jonas Shoeley’s slides

Tidyverse

The most influential R developer

Hadley Wickham

tidyverse

tidy data

On pipes

Examples and exercises

Please follow me on the R script "tidy.R"

Visualizing data with ggplot2

A bit more motivation

More

Plotting systems in R?

“The winner takes it all”

Strength of base plotting system

What makes ggplot2 special?

Extremely big and helpful community

ggplot2 resources

One last hint before we start coding

ggplot2 show

Please follow me on “ggplot2.R”

Please follow me on the R script `"tidy.R"`

Visualizing data with `ggplot2`

Strength of `base` plotting system

What makes `ggplot2` special?

`ggplot2` show

Please follow me on “`ggplot2.R`”